80 results about "Grid clustering" patented technology

The invention provides an LEACH protocol improvement method capable of lowering energy consumption. A basic system comprises normal nodes, cluster header nodes, forwarding nodes and a base station. The specific cycle operation steps comprise network topology determination, initialization, selection of the cluster header nodes, cluster forming, data transmission, selection of the cluster header nodes and cluster forming. According to the defects of an LEACH protocol, improvement is conducted from the aspects of selection of cluster heads, cluster forming and a data transmission mode. In selection of the cluster heads, the comprehensive influences of the energy of the nodes, the distance to the base station and density are considered, the burdens of the cluster heads are balanced, and distribution and energy consumption of the selected cluster header nodes are made balanced in a network. In cluster forming, the mode of grid clustering is used, the nodes calculate the two-dimensional values of grids where the nodes are placed by themselves, and the effectiveness of node energy is guaranteed. In data transmission mode, a routing mode with single-hop and multi-hop combined is used according to the distance of the cluster header nodes and the base station, the energy consumption of network nodes is lowered by the purpose that the energy utilization rate is highest.

The invention discloses a method for fusing sensor grid data based on grid clustering, which comprises the following steps: allocating tasks to all sensor nodes of each wireless sensor network, and then running an adaptive genetic algorithm based on the grid clustering in each wireless sensor network; according to perceptive data of all the sensor nodes, finding out an optimal sensor node and then finding out an optimal fusing routing with the minimum communication cost; primarily fusing data by a mobile agent along the optimal fusing routing and then sending the primarily fused result to nodes of convergent points of the network; and adopting an adaptive weighted fusion algorithm by a grid layer to carry out deep fusion on the data so as to obtain the final fused result which is providedfor a user for making decisions. The method can be applied to a sensor grid environment and is used to process the data sufficiently so as to obtain the good fused result. The grid is used for deeplyfusing the data, thus reducing energy consumption of the wireless sensor networks and prolonging the service life of sensor grids.

The invention belongs to the technical field of wireless network protocols, in particular relates to an opportunistic routing method based on grid clustering in a wireless sensor network. The invention adopts a grid-type clustering method and the opportunistic routing algorithm to fully take advantages of the energy effectiveness of a clustering topology control method and the advantage of an opportunistic routing in the aspect of end-to-end throughput, thereby ensuring that the sensor network achieves effective optimization in the aspects of data throughput, transmission reliability, and the like on the premise of energy saving.

The invention relates to a grid-clustering-based RSSI (Receive Signal Strength Indicator) indoor positioning algorithm which comprises the following steps: S1, placing an RFID (Radio Frequency Identification) fixed reader, randomly placing an active tag as a to-be-detected tag, and determining an RFID positioning system model; S2, repeatedly measuring the RSSI value from each reader to the to-be-detected tag at different moments, recording the measured value each time, and calculating the distance from the to-be-detected tag to each reader according to the acquired RSSI value; S3, acquiring aplurality of 'estimation tags' by using a trilateration algorithm, solving the mean of the coordinates of the 'estimation tags' as a 'target tag'; S4, defining a search region by taking the 'target tag' as a center, and dividing the region into grids of the same size; S5, screening the grid containing the most 'estimation tags' by utilizing a grid clustering algorithm, and selecting the 'estimation tags' in the grid as reference tags; S6, solving the centroid of the reference tags as the estimated value of the to-be-determined tag coordinate according to a weighted centroid algorithm.

The invention provides a telephone traffic distribution calculation method based on neighbor cell field intensity information and a system thereof. The method comprises the following steps: A, carrying out meshing on an area to be calculated of telephone traffic distribution, obtaining average filed intensity lists of different antennas received by each grid; B, carrying out clustering on grids in the area to be calculated of telephone traffic distribution, and forming primary service area grid clustering division of different antennas; C, collecting a measurement data report and total telephone traffic data generated by each antenna within a period in the area to be calculated of telephone traffic distribution; D, by utilizing the measurement data report and total telephone traffic data of each antenna, the primary service area grid clustering division, and the average filed intensity list of different antennas received by each grid, calculating telephone traffic distribution of each grid in the area to be calculated. According to the method and the system in the invention, service quality of a mobile communication network is substantially raised, and work efficiency of communication network programming and optimizing personnel are raised.

The invention provides an on-board LiDAR point cloud road information extraction method and device. The method comprises the following steps that point cloud data of a to-be-detected road is filteredto obtain ground point cloud data; a two-dimensional net is established for ground point cloud data; first average elevation value for each grid, a highest elevation, a lowest elevation for and secondaverage elevation value of all adjacent grids are determined, existence of road point grids in all the adjacent grids is determined according to values; when the road point grids exist, grid clustering is carried out for all road point grids to determine road point grid clusters; and according to the size of each road point grid cluster, whether the point cloud data in the road point grid clusterare road point cloud data is determined. According to the technical scheme, a standard spatial grid is established for the point cloud, location of the road is determined by the spatial distributioncharacteristics and the relationship between elevation differences; and the grids with roads are determined according to the grid cluster growth mode to achieve the aim of road extraction.

The invention requests to protect a grid clustering algorithm based on a density peak value, and can efficiently processing large-scale data. The algorithm comprises the following steps that: firstly, pelletizing an N-dimensional space into nonintersecting rectangular grid unit, and then, carrying out statistics on the information of unit space; utilizing a thought that a density peak value clustering is used for searching a center point to determine a center unit, i.e., surrounding the center grid unit by certain data units with low local density, and forming a large distance with the grid unit with high self local density; and finally, combining the grid units near the center grid unit so as to obtain a clustering result. A situation experiment result on an UCI (University of CaliforniaIrvine) artificial data set indicates that the algorithm can quickly obtain a clustering center, the clustering problem of large-scale data can be effectively processed, and the algorithm is high in efficiency. Compared with an original density peak value clustering algorithm, the algorithm is characterized in that time loss on different datasets is lowered by 10-100 times and accuracy loss is maintained at 5-8%.

The invention discloses a system and a method for mining a workplace-residence position tag based on mobile communication signaling big data. The key point of the technical scheme is as follows: the system comprises a user individual activity sequence establishment unit used for establishing user individual activity sequences, a grid generation unit used for dividing position information of a cell into uniform grids, a grid cluster analysis calculation unit used for calculating occupied dense grid clusters in appointed time frame by calling a grid cluster analysis calculation algorithm, a density cluster analysis calculation unit used for calculating mass centers of the occupied dense grid clusters by calling a density cluster analysis calculation algorithm, and a workplace-residence position tag generation unit for generating the workplace-residence position tag of the user. The system and the method have the advantages as follows: 1) any assumption on the user behavior sequence data distribution is not required and the obtained result through analysis is not related to data record processing sequence, so that relatively high objective degree is achieved; 2) sub spaces with cluster probability are found automatically; and 3) the complexity of data ETL is lowered.

The invention discloses a grid clustering analysis-based violence behavior detection method. The method comprises the following steps of: S1, reading a video streaming from camera equipment or an existing video file, and obtaining source picture frames in the video streaming; S2, preprocessing the source picture frames; S3, detecting human bodies in the preprocessed source picture frames and framing out the detected human bodies; S4, obtaining optical flow feature points of the source picture frame and carrying out optical flow tracing on the optical flow feature points; S5, carrying out clustering analysis on the optical flow feature points to obtain moving objects in the image; S5, analyzing moving conditions of the moving objects; and S7, carrying out statistic on a degree of polymerization change trend of the moving objects in continuous image frames so as to judge the detection of violence incidents. According to the method, the degree of polymerization change trend among a plurality of moving objects is calculated through video image information, and human body detection is carried out to judge whether violence incidents occur or not.

The invention discloses a traffic jam point discovery method based on Spark cluster parallel computing. The invention discloses a trajectory big data mining method, relates to the technical field of trajectory big data mining, solves the technical problem of how to quickly cluster massive trajectory data and find a traffic congestion source of an urban hot region, and comprises the following steps: (1) preprocessing the massive data, including complementing errors between trajectory points and eliminating trajectory point redundancy in a certain region for a period of time; (2) adopting a grid-means clustering algorithm to carry out clustering to obtain a plurality of target data samples; (3) clustering by adopting a neighborhood maximum density grid clustering algorithm to obtain an urbanhot traffic road network model; and (4) calculating the adjacent grid density difference by adopting a neighborhood density difference algorithm to obtain a traffic jam source, and storing the obtained result in a Spark memory. According to the method, mass data can be quickly clustered to obtain an urban traffic network model, and quick discovery of a traffic jam source region is realized.

A system and method for automatically generating a computation mesh for use with an analytical tool, the computation mesh having a plurality of ξ-grid lines and η-grid lines intersecting at grid points positioned with respect to an inner boundary and an outer boundary. The system and method include one or more mesh equations having one or more source terms that include: a grid clustering component based on a Jacobian scaling parameter, a source decay parameter, and one or more first point distance parameters, and a cell shape modifying source component based on one or more source parameters selected from the group consisting of a smoothing source parameter, an area source parameter, an orthagonality source parameter, and any combinations thereof.

The invention relates to a grid division processing method and device based on a city map, computer equipment and a storage medium in the field of artificial intelligence. The method comprises the following steps: acquiring a city identifier and a corresponding city map, and dividing the city map to obtain a plurality of regional grids; acquiring corresponding user information according to the city identifier; dividing each user into a corresponding regional grid according to the user information; wherein the user information comprises geographic coordinates and uploading time; identifying thearea type of each area grid by utilizing the geographic coordinates of each area grid in the preset time period and the corresponding uploading time; and acquiring a plurality of regional grids in apreset range of the city map, performing regional grid clustering according to the distances among the regional grids and the regional types to obtain an activity region of the user, and marking the activity region in the city map. According to the method, the city map is divided, so that the user activity area can be quickly and effectively identified.

The invention discloses a rapid night vehicle detection method applied to a self-adaptive high beam, and the method comprises the steps: step 1, enabling an image collection module to collect a road traffic image in front of a vehicle, and transmitting image data information to an image processing module; step 2, processing the image data information by an image processing module, and judging a suspected vehicle lamp region by adopting a grid clustering algorithm; step 3, determining a halo range of a suspected vehicle lamp region by adopting a corrosion algorithm, calculating a halo color through a rapid algorithm, and judging a headlamp and a tail lamp; step 4, conducting pairing according to the geometrical relationship, recognizing vehicles, calculating vehicle coordinate position information and achieving night vehicle detection; and step 5, enabling a data transmission module to transmit the vehicle coordinate information calculated by the image processing module to the high beamcontrol module. The final vehicle lamp information obtained through image processing can serve as the control basis of the self-adaptive high beam lamp and can also provide support for other modulesneeding the vehicle lamp information.

The invention discloses a dynamic grid generation method. The dynamic grid generation method comprises the following steps: taking a sealed region formed by maximum and minimum longitude and latitude lines of a current display region as a reference region of the current display region; finding out a reference point closest to any point of the reference region as a closest reference point, wherein the closest reference point is found from a pre-stored reference point database or is instantly generated from the known reference points; and taking the closest reference point as a starting point for extending in the longitude and latitude directions to generate a minimum dynamic grid which is composed of a plurality of unit grids and completely covers the current display region. The invention further discloses a dynamic grid generation device, and a polymerization method and device based on the dynamic grid. With the adoption of the embodiment of the invention, the problems in a grid clustering algorithm that a lot of data need to be stored, the data processing speed is low and the polymerization errors are great can be effectively solved.

The invention discloses a dynamic grid generating method. The method comprises the following steps of making an enclosed area formed by the largest and smallest latitude lines and longitude lines on a current display area serve as the reference area of the current display area; finding out a reference point nearest to any point on the reference area as a nearest reference point; judging whether to generate a current display area reference point according to the position relationship of the nearest reference point and the reference area; if yes, then extending towards the latitude and longitude directions with the current display area reference point as a starting point to generate a minimum dynamic grid composed of a plurality of unit grids and fully covering the current display area; and if no, then creating a minimum dynamic grid with the nearest reference point as the starting point. The invention further discloses a dynamic grid generating apparatus, a clustering method and a clustering apparatus based on the dynamic grid. By means of the method, the problems of a grid-clustering algorithm which requires storage of large amounts of data and results in a low data processing speed and great clustering errors are solved.

The invention discloses a dynamic clustering algorithm for geographic grids. The method includes the steps that firstly, the maximum longitude and latitude and the minimum longitude and latitude are found out in an area, and then according to the step length, grid division is performed on the area between the maximum longitude and latitude and the minimum longitude and latitude; serial numbers of the grids where all points are located are calculated, the points in the same grid are clustered, and the gravity center of clustering is calculated; then the grids are moved in the vertical direction and the horizontal direction, the moving lengths are kept uniform, the frequencies of moving in the directions mutually symmetrical are kept uniform, the sequence of the moving directions is not limited, after moving each time, the points, meeting clustering conditions, in the grids are clustered, and all the clustered points form a final clustered result. According to the method, the problems that a traditional clustering algorithm based on the grids is low in precision and efficiency are solved, the grids can be drawn dynamically through the method, the granularity can be freely controlled, flexibility is high, and the speed is high.

The invention provides a multi-hypothesis multi-target track initiation method based on grid clustering. The multi-hypothesis multi-target track initiation method comprises the steps of obtaining a trace point set and mapping the trace point set into a grid; determining boundary points of the current high-density grid; forming a trace point cluster based on boundary points of the high-density grid; performing clustering processing on each trace point cluster to generate a cluster; establishing an association hypothesis of a target and measurement in the cluster; and obtaining an optimal hypothesis according to the hypothesis probability obtained through calculation, calculating track scores, sorting the track scores from high to low, deleting low-score tracks through backtracking so as toconfirm track generation, and updating target state estimation of the tracks. According to the method, a plurality of target tracks can be initiated, the problem of multiple false tracks in a high-density clutter environment is solved, and engineering application under a high-density clutter condition is facilitated.

The invention discloses a method for intelligently making a peak shaving and load shifting strategy for power customers, and the method comprises the steps: carrying out the clustering of load curves of a regional power grid in a certain time through employing a two-step clustering algorithm, and obtaining a regional power grid clustering result, wherein the regional power grid clustering result comprises a clustering center curve and clustering details; carrying out the dividing of the peak and valley of the load curves of the regional power grid, and carrying out the matching of a power customer curve at a corresponding date through combing with the regional power grid clustering result; carrying out the clustering of the power customer load curve of a certain type of regional power grids in the same period at the peak and valley through employing the two-step clustering algorithm, obtaining the clustering result of the power customer, wherein the clustering result comprises a clustering center curve and clustering details; and constructing a peak shaving and load shifting index of the power customer according to the normalized mean value of the power customer clustering center curves of the regional power grid and the similarity between the regional power grid and the corresponding power customer clustering center curve. The method is high in practicality, and irons out a defect that the recommended peak shaving and load shifting strategy of the power customer does not accord with the actual condition because only the curve shape or load absolute value is considered.

The invention discloses a multi-extended target tracking method under space-time neighbor target detection and grid clustering measurement division, and mainly solves the problem of multi-extended target tracking under measurement division. The method comprises the following steps of firstly, according to a space-time correlation principle, centralizing current measurement into survival expansionand newly-expanded target measurement; secondly, dividing survival extension target measurement by using an FCM algorithm; carrying out dynamic grid processing on the measurement space of the newly-generated extended target, removing clutter measurement according to the grid density and the look-up distance, and carrying out grid clustering method division; Finally, utilizing a GM-PHD filter and an ET-GM-PHD filter to estimate the states of the survival and new multi-expansion target states to obtain the motion trails of the survival and new multi-expansion target states. The method can accurately track multiple extended targets, and particularly has the advantages of accurate measurement division, accurate target number estimation, high tracking precision, good real-time performance and the like at the target intersection moment.

The invention discloses a video image stabilization motion estimation method based on reference frame optimization and foreground and background separation, which is used for solving the problem thatthe effect after image stabilization processing is influenced by inaccurate motion vector estimation. The method comprises steps of firstly, carrying out reference frame selection optimization by taking adjacent frame priority as a principle, and then carrying out clustering and preliminary screening on feature points in video frames in a grid clustering and density clustering algorithm fusion mode; solving an optimal homography matrix by using a random sampling consistency algorithm, and solving the distance between the feature point pairs by combining perspective projection transformation; and removing the feature point pairs which do not meet a preset condition according to a distance criterion to obtain purified feature point pairs; and finally, using the obtained accurate background feature point pairs for calculating a global motion vector. According to the video image stabilization motion estimation method disclosed by the invention, the influence of the foreground on motion estimation is effectively removed, the estimated global motion vector precision is improved, the image stabilization restoration effect is improved, and the method can be widely applied to image stabilization restoration of videos shot by different shooting devices.

The embodiment of the invention provides a common-mode antenna feeder optimization method and system based on adaboost weighting and third-order clustering. The method comprises the steps of: fusing drive test data, first data and position information in any one network, and carrying out grid clustering on the fused data so as to obtain a data set of any one network; respectively determining an optimal weight corresponding to the data set of any network by adopting an AdaBoost algorithm according to the respectively set initial weights corresponding to the data set of any network, and obtaining a second data set according to the optimal weight corresponding to the data set of any network and the data set of any network; selecting second sampling points of which RSRP or SINR in the second data set is smaller than a preset threshold value, and clustering each sampling point in the second sampling points with the position information to generate a difference geographical position clustering set; and calculating included angles between the horizontal lobe normal of an antenna in a first common-mode cell and the difference point geographic position clustering set and between the vertical lobe normal of the antenna in the first common-mode cell and the difference point geographic position clustering set according to the difference point geographic position clustering set and the antenna feed value of the first common-mode cell.

The invention relates to a parallel clustering method, in particular to a parallel clustering method for processing large geographical grid data and belongs to the field of parallel clustering for processing the large geographical grid data. The problems that because geographical grids are quite enormous and exceed a stand-alone loading limit, a system can not operate or operation efficiency is quite low, and all progress computing results interchange mutually when multiple computers perform parallel clustering are solved. The method includes the steps that step1, the number of nodes is calculated, and each computing progress is numbered; step2, the data are transmitted to computing progresses, and the numbers of the computing progresses are IDs; step3, M sets of clustering solution initial values are generated; step4, a clustering vector central table is sent to the computing progresses; step5, a management progress controls an iteration solution progress; step6, content of the row, with the highest value, of a field 4 is sent to all the computing progresses; step7, grid clustering results are written into a geographical grid data file. In this way, the parallel clustering method for processing the large geographical grid data is implemented through the steps.

The invention discloses an improved method for solving a fan power curve parameter model based on a genetic algorithm. According to the method, each parameter in the fan power curve parameter model is determined through fan operation data in an operational process of wind power units. But in a process for solving the parameters of the model by using the genetic algorithm, the fitness of all individuals in a population is calculated in each iterative process, and when the data volume of the fan operation data is increased, the calculation amount of a fitness function and an evaluation function of the population is increased accordingly and the space complexity and time complexity of a program are also increased. Therefore, the fan operation data is subjected to grid-based clustering and weight assignment first and then each parameter in the parameter model is determined by applying the genetic algorithm, in the method. By clustering the fan operation data, the increment of the calculation amount of the fitness function and the evaluation function caused by the increment of the data volume of the fan operation data can be avoided; and by assigning weights to clustering points, the model can adapt to most original data points and the credibility of the model is enhanced.

The invention discloses an optimized method of variable k coverage for a two-dimensional intrusion monitoring region. The method comprises the following steps: step (1), abstracting an intrusion object as a point and constructing a two-dimensional monitoring region coverage model of single point intrusion; and step (2), meshing a two-dimensional monitoring region according to the two-dimensional monitoring region coverage model of single point intrusion and a grid clustering algorithm, performing clustering according to the coverage of each grid cell after meshing, movably deploying sensor nodes in the two-dimensional monitoring region according to the comparison of the current coverage with the desired coverage, and finally realizing the variable k coverage of the two-dimensional monitoring region. According to the method, the nodes are re-deployed to realize flexibly variable coverage of different regions at different times; in the case of limited sensor nodes, the effects of the nodes are developed thoroughly to ensure multiple coverage of a key region and also satisfy the coverage requirement of a common region; as a result, effective monitoring on the intrusion object can be realized at the cost of consuming less energy.

The invention discloses a clustering method based on an adjacent grid search strategy, and the method comprises the steps: firstly carrying out grid division of original data: enabling an original data set to be divided into a limited number of cells through a multi-dimensional space grid, and carrying out the denoising processing when necessary; secondly, performing grid clustering on the divided data: processing the denoised grids by utilizing a halo threshold value, and dividing the denoised grids into halo cell elements and core cell elements; establishing an adjacent grid operator for quickly searching an adjacent cell element of one cell element; achieving a clustering process through two steps of core cell element clustering and halo cell element division, dividing all core cell elements into a plurality of class clusters through a traversal algorithm, and dividing halo cell elements into existing class clusters based on cell element distances; and finally, clustering optimization is carried out according to data characteristics and user requirements. Compared with the prior art, the method has the advantages that a new clustering method can be provided for rapidly increasing the dimension of the sample set, and class clusters with complex boundary shapes can be effectively identified.